First, I have to confess that I am not an expert in database
system ( I can't tell whether we are talking about an object
repository, because I do not know what it is ).
Concerning my software, however I can explain a bit more:
I store serialized Perl objects in a DB_File database. These
objects are stored in a filesystem like tree; that is objects may contain other objects, such as directories
contain files.
There is an API to access objects; I have been working hard
with Psh and Apache+mod_perl, but any other Perl program can
access the object filesystem ( I have written other scripts
to access the object filesystem, such as permanent daemons
and cron jobs ). The thing is that as an administrator, I
prefer the Psh command line to fix broken objects, but users
have only access through the HTTP server.
A great part of the concepts I implement, are taken from
the Unix world, I just transposed them to the Perl object
world.
I have also written a bunch of other modules ( not in the
tar you can get from geocities ):
Objects whose attributes are indexed ( either in an SQL
database or in a BTREE DB_File ).
Objects that may contain a large number of other objects
with fast access to subobjects.